KR101265956B1 - System and method for restoration image based block of image - Google Patents

Abstract

A block-based image reconstruction system and method are disclosed. The block-based image reconstruction system divides the image into at least one region and a boundary processor that processes the boundary of the image by performing color shifting in the boundary region of the image, and includes a block included in the divided region. And a reconstruction parameter extracting unit for extracting reconstruction parameters for each image and a reconstructing unit for reconstructing the image by applying block-based transform region filtering according to the reconstruction parameter.

Scaler, resolution, reconstruction parameters, Fourier transform region filtering, wavelet transform region filtering

Description

Block-based Image Restoration System and Method {SYSTEM AND METHOD FOR RESTORATION IMAGE BASED BLOCK OF IMAGE}

The present invention relates to a block-based image restoration system and method, and more particularly, to a block-based image restoration system and method capable of restoring high frequency region loss of an image generated when improving the resolution of an image and emphasizing the detail of the image. It is about. The present invention can be applied to display devices such as CRTs, PDPs, LCDs, and projection TVs. In particular, the present invention can be applied to ultra-high resolution display devices such as UD (Ultra Definition) TVs and digital cinemas.

Display apparatuses for video have been developed with the aim of increasing the size of a screen and expressing a high resolution image. In this situation, the image content also needs to be improved in low resolution to high resolution in order to be displayed on a larger screen. In particular, according to the development of the image technology, a UD capable of using an image scaler to display an image representing a resolution of SD (Standard Definition -720 * 480) or HD (High Deifintion-1920 * 1080) in a high resolution image device. There is a need to convert an image having a resolution of (Ultra Definition-7680 * 4320).

In this case, an image scaler is generally used to convert a low resolution image into a high resolution image. However, since the image scaler improves the resolution of the image based on interpolation, loss occurs in the high frequency region of the image (particularly, the edge region of the object), and there is a problem in that the detail of the image is insufficient. In particular, the loss in the high frequency region of the image appears as a blur occurring at the boundary of the object. The blur may be a result of a combination of multiple blurs, such as those generated when an image is acquired and those caused by interpolation performed by the scaler.

An image reconstruction process is required to remove the blur in the image due to the loss of the high frequency region. In the conventional image reconstruction apparatus, since the high frequency component of the image is restored by removing the blur of the image without considering the feature of each region of the image, the image reconstruction apparatus has a lack of visual unity when viewing the image as a whole. In particular, in the case of an ultra high resolution image, the characteristics of the image are different for each region because of the large screen size. This is because the image data itself exhibits statistical characteristics that change spatially.

Therefore, there is an urgent need for a method capable of restoring loss of a high frequency region of an image while maintaining visual unification of the image in consideration of the entire information and detailed information of the image.

An image reconstruction system according to an embodiment of the present invention divides the image into at least one region and a boundary processor that processes the boundary of the image by performing color shifting in the boundary region of the image, and divides the divided image. And a reconstruction parameter extracting unit extracting reconstruction parameters for each block included in the area, and an image reconstructing unit performing image reconstruction by applying block-based transform region filtering according to the reconstruction parameter.

According to an aspect of the present invention, the boundary processor may remove the ringing component of the boundary region of the image by shifting the object color and the background color from the boundary of the object color and the background color of the image to respective representative colors. have.

Therefore, according to the present invention, the image reconstruction can be performed without removing the deterioration of the image quality by removing the ringing component of the boundary region of the image through the boundary processing. In addition, since the ringing component is removed, the sharpness of the image may be improved when the image is restored, and the performance of the image restoration may be improved as a whole.

According to an aspect of the present invention, the reconstruction parameter extractor may extract a blur value and a noise removal threshold of each block by using an analysis result of the spatial activity of each block.

According to an aspect of the present invention, the reconstruction parameter extractor may extract reconstruction parameters based on global reconstruction information for the entire image and local reconstruction information for each of the blocks.

Therefore, according to the present invention, by reconstructing the reconstruction parameter for each block, image reconstruction can be performed more quickly, and generation of blocking artifacts can be prevented by considering global reconstruction information for the entire image.

According to an aspect of the present invention, the image reconstruction unit may apply Fourier transform region filtering and wavelet transform region filtering to each of the blocks using the reconstruction parameter.

Therefore, according to the present invention, blur caused by resolution of an image may be effectively removed through Fourier transform region filtering, noise may be suppressed through wavelet transform region filtering, and the sharpness of the entire image may be improved.

An image restoration method according to an embodiment of the present invention comprises the steps of processing the boundary of the image by performing color shifting (color shifting) in the boundary region of the image, dividing the image into at least one region, the divided region Extracting a reconstruction parameter for each block included in the block and performing image reconstruction by applying block-based transform region filtering according to the reconstruction parameter.

Hereinafter, with reference to the contents described in the accompanying drawings will be described in detail an embodiment according to the present invention. However, the present invention is not limited to or limited by the embodiments. Like reference symbols in the drawings denote like elements.

1 is a block diagram illustrating a specific configuration of an image reconstruction system according to an embodiment of the present invention.

Referring to FIG. 1, the image restoration system 101 may include an image scaler 102, a boundary processor 103, a restoration parameter extractor 104, and an image restoration 105.

The image scaler 102 may improve the resolution of the image by upsampling the image and interpolating the upsampled image. In particular, when improving the resolution of a low resolution image to a high resolution image, the image scaler 102 may be used.

For example, the image scaler 102 may upsample an image by using zero data and interpolate the upsampled image through a low band filter. In detail, the image scaler 102 may insert zero data to increase the size of the image and remove unnecessary components by using interpolation. That is, the image scaler 102 may improve the resolution of the image based on interpolation.

However, when the image scaler 102 is used alone to improve the resolution of the image, blur may occur for the entire image. In this case, the generated blur may be a result of a combination of multiple blurs, such as blur generated when acquiring an image and blur generated due to interpolation performed by the image scaler 102. Here, the generated blur may be represented by a point spread function of the blur. Since the point spread function represents the shape of the low pass filter, the noise component is suppressed as the blur of the image increases. In this case, since the noise component is independent of the image signal, the noise dispersion value may be used to predict the blur of the image.

The image scaler 102 may use a low pass filter (LPF) when interpolating. However, since the low-band filter is used for interpolation, high frequency components (eg, edge regions) inherent to the image may be lost, and the sharpness of the image may decrease according to the loss of the edge region. In particular, when the image is improved to an ultra high resolution image such as an ultra definition (UD), the loss of the high frequency component may appear even more because the resolution enhancement magnification is large. Accordingly, the image reconstruction system according to an embodiment of the present invention may improve the sharpness of an image by restoring a loss of a high frequency component that is inevitably generated in the resolution improvement process of an existing image.

The boundary processor 103 may process the boundary of the image by performing color shifting on the boundary region of the image. In this case, the image refers to an image having an improved resolution through the image scaler 102. For example, the boundary processor 103 may remove a ringing component of the boundary region of the image by increasing a signal bandwidth of one of an object color and a background color of the image. In this case, the ringing component is generated between the edge of the object and the edge of the background, and may appear due to blur caused during the resolution improvement of the image performed by the image scaler 102.

The image reconstruction system according to an embodiment of the present invention may perform image reconstruction by emphasizing a high frequency component of an image having an improved resolution. At this time, when the ringing component is present in the boundary region of the image, when the high frequency component is emphasized, the ringing component is also improved, which may cause degradation of image quality. Therefore, according to an embodiment of the present invention, the boundary processor 103 may improve the performance of image reconstruction by performing boundary processing to remove ringing components existing at the boundary of the image. That is, the boundary processor 103 may remove the ringing component by designating the color of the ringing component as the object color or the background color. Specific operations of the boundary processor 103 will be described in detail with reference to FIGS. 2 and 3.

The reconstruction parameter extractor 104 may divide an image into at least one area and extract reconstruction parameters for each block included in the divided area. The recovery parameters are extracted independently for each divided region, and the number of divided regions is not limited. If the image is enhanced to UD resolution, since the image data is very large, it may be inefficient for the restoration parameter extractor 104 to extract the restoration parameters for the entire image. Therefore, the restoration parameter extracting unit 104 may extract the restoration parameters for each block included in the divided region.

For example, the reconstruction parameter extractor 104 may extract the blur value and the noise removal threshold of each block by using the analysis result of the spatial activity of each block. In general, since a large image such as an ultra-high resolution image exhibits different characteristics for each region of the image, it is more effective to extract different parameters for each region than to extract the same restoration parameter for the entire image. Can be done. Accordingly, the restoration parameter extracting unit 104 according to an embodiment of the present invention may extract the blur value and the noise removal threshold of each block.

In this case, a blocking artifact, which is a block boundary phenomenon, may occur due to an excessive difference in reconstruction parameters between blocks in an image having a large resolution. Thus, as an example, the reconstruction parameter extractor 104 may extract reconstruction parameters based on global reconstruction information for the entire image and local reconstruction information for each block. The restoration parameter extracting unit 104 will be described in detail with reference to FIGS. 4 and 5.

The image reconstructor 105 may reconstruct an image having improved resolution by applying block-based transform region filtering based on the extracted reconstruction parameter. For example, the image reconstructor 105 may apply Fourier transformation domain filtering and wavelet transformation domain filtering to each block by using the reconstruction parameters.

For example, the image reconstructor 105 may remove a blur of an image by applying Fourier transform region filtering. That is, the image reconstructor 105 may effectively remove the blur by inversely transforming a function on the blur value through Fourier transform region filtering. However, when Fourier transform region filtering is applied, it is effective to remove blur, but noise boost may occur.

Accordingly, according to an embodiment of the present invention, the image reconstructor 105 may apply wavelet transform region filtering together with Fourier transform region filtering to remove noise and improve sharpness of an image having an improved resolution.

However, in the case of an image having a large resolution, there is a problem in terms of processing speed when serially applying transform region filtering. Thus, as an example, the image reconstructor 105 may apply Fourier transform region filtering using a blur value and wavelet transform region filtering using a noise removal threshold in parallel in consideration of the filtering process for each divided region. You can quickly restore the image. Fourier transform region filtering and wavelet transform region filtering are described in detail with reference to FIGS. 6 and 7, respectively.

2 is a view for explaining the effect of the boundary treatment when restoring a high frequency component according to an embodiment of the present invention. Specifically, FIG. 2 is a diagram for explaining the effects of boundary processing by comparing the effect of the high frequency component on the image in which the ringing component is present and the image in which the ringing component is removed.

The graph 201 shows a signal of an image passing through the image scaler 102. As mentioned above, the image scaler 102 may cause blur in the boundary region of the image because the image scaler 102 improves the resolution of the image based on interpolation. As a result, a ringing component between the object and the background may occur due to blur occurring in the boundary region of the image. According to the graph 201, a ringing component is present at the boundary of the image when an overshoot exists in a portion having a large brightness value and an undershoot exists in a portion having a low brightness value. It can be seen.

The graph 202 shows an image signal when a high frequency component is emphasized for an image having a ringing component in the boundary region. That is, if the high frequency component is emphasized as it is for the image having the ringing component in the boundary region through image reconstruction, the ringing component is also emphasized, such that the image quality may deteriorate as shown in the graph 202.

The graph 203 shows an image signal from which a ringing component is removed through boundary processing. For example, the boundary processing unit 103 may remove the ringing component by performing boundary processing through color shifting. The graph 203 shows that the overshoot or undershoot, which means a ringing component, does not appear in the video signal.

The graph 204 shows an image signal when the high frequency component is emphasized with respect to the edge processed image. Unlike the graph 202, the graph 204 can be seen that no deterioration in image quality occurs. In addition, referring to the graph 204, it can be seen that the difference between the region having a large brightness value and the region having a small brightness value is further increased, so that the contrast of the boundary region of the image is improved, and the sharpness of the boundary portion of the image is further increased. Can be. As a result, according to an embodiment of the present invention, when the high frequency component is emphasized with respect to an image from which the ringing component is removed through boundary processing, an image with increased sharpness of the image may be obtained while blurring of the image boundary is eliminated. .

3 is a view for explaining the operation of the boundary processor of the image restoration system according to an embodiment of the present invention. In detail, FIG. 3 is a diagram illustrating a process of removing the ringing component by the boundary processor 103 of the image reconstruction system 101.

Reference numeral 301 denotes that a ringing component exists in the boundary region of the image. As mentioned above, the ringing component means blur of an image generated when the image scaler 102 improves the resolution of the image.

Reference numeral 302 denotes that the boundary processing unit 103 processes the boundary of the image by performing color shifting in the boundary region of the image. In this case, the boundary processor 103 may remove the ringing component of the boundary region of the image by shifting the object color and the background color from the boundary between the object color and the background color of the image to each representative color.

That is, since the ringing component is present on both sides of the boundary between the object and the background of the image, the boundary processing unit 103 changes the signal bandwidth through the process of shifting the color of the object image around the boundary and the color of the background image to respective representative colors. It can be increased to remove the ringing component present in the boundary region of the image. Then, as shown by reference numeral 303, it can be seen that the ringing component is removed from the boundary region of the image through color shifting. When the image reconstruction is performed after the boundary processing process of removing the ringing component, the resultant image may be derived as the sharpness is improved and the amplification of the ringing component is minimized.

4 is a diagram for describing a process of performing block-based image reconstruction according to an embodiment of the present invention.

In detail, as illustrated in FIG. 4, the reconstruction parameter extractor 104 may divide the entire image into at least one area and extract reconstruction parameters for each block included in the divided area. At this time, the restoration parameter extracting unit 104 may extract the restoration parameters independently according to the divided region. In addition, the image reconstructor 105 may perform image reconstruction by applying block-based transform region filtering according to the reconstruction parameter. At this time, the number of regions for dividing the entire image is not limited.

For example, since an ultra high resolution image representing a UD resolution has a large amount of data, it is very complicated and inefficient to perform image restoration on the entire image. Therefore, according to an embodiment of the present invention, the restoration parameter extracting unit 104 and the image restoration unit 105 may perform image restoration on a block basis. Referring to FIG. 4, the entire image is divided into four regions, and a restoration parameter is extracted from left to right with respect to blocks included in each region, and an image restoration is performed. The extraction direction of the reconstruction parameter may be arbitrary.

However, as mentioned above, when the reconstruction parameter for each block is extracted, block artifacts may occur when the difference between the blocks is large. Accordingly, the block parameter extractor 104 may extract the restoration parameter based on the global restoration information for the entire image and the local restoration information for each block. In detail, the reconstruction parameter extractor 104 may extract excessive block parameters based on global reconstruction information for the entire image as a guideline, thereby preventing excessive differences in reconstruction parameters between blocks.

5 is a diagram for describing a detailed operation of a block parameter extracting unit of an image restoration system according to an embodiment of the present invention.

As mentioned above, the block parameter extractor 501 may segment an image into at least one region and extract a reconstruction parameter for each block included in the divided region. For example, referring to FIG. 5, the block parameter extractor 501 may perform a spatial activity analysis process 502 and a restoration parameter extraction process 503 of each block. In this case, the block parameter may include a blur value and a noise removal threshold of each block. In one example, the spatial activity may be expressed through Equation 1 below. In this case, the block parameter extractor 501 may use the size of the block included in the local restoration information 504.

,

는 각각 x축, y축 방향으로의 그레디언트(gradient) 값을 나타낸다. a는 각 블록의 공간적 활성도를 나타내고, k는 블록의 인덱스를 나타낸다. here,

,

Denotes gradient values in the x- and y-axis directions, respectively. a represents the spatial activity of each block, and k represents the index of the block.

The reconstruction parameter extractor 501 may prevent the occurrence of blocking artifacts by independently extracting reconstruction parameters according to the divided regions of the image in consideration of the global reconstruction information 505 together with the local reconstruction information 504. The restoration parameter extractor 501 may extract the blur value and the noise removal threshold of each block, which are restoration parameters, in consideration of the global restoration information 505 together with the local restoration information 504. For example, the blur value and the noise removing threshold may be expressed through Equation 2 below.

와

는 복원 파라미터인 블록 각각의 블러값과 노이즈 제거 임계값을 평균한 값으로, 글로벌 복원 정보(505)에 해당한다. 이 때, 글로벌 복원 정보(505)는 영상 스케일러부(102)가 영상의 해상도를 향상시킬 때 적용한 배율 정보에 따라 결정될 수 있다. 이는 영상 스케일러부(102)의 배율이 일정할 경우, 블러 되는 정도와 노이즈 부스트 정도가 평균적으로 일정함을 이용한 것이다.here,

Wow

Is a value obtained by averaging the blur value and the noise removal threshold value of each block, which is a restoration parameter, and corresponds to the global restoration information 505. In this case, the global reconstruction information 505 may be determined according to the magnification information applied when the image scaler 102 improves the resolution of the image. This means that when the magnification of the image scaler 102 is constant, the degree of blurring and the degree of noise boost are constant on average.

In one example, the average value of the spatial activity is most accurate to calculate within the current frame of the image, but if the movement of the object included in the image for the hardware implementation is not very large, it can be calculated within the immediately preceding frame.

와

는 step-size이고,

와

는 임의의 매핑 함수이며, 각각 로컬 복원 정보(504)에 해당한다. 또한,

는 블록의 공간적 활성도이고,

는 블록 각각의 공간적 활성도를 평균한 값이다.And,

Wow

Is step-size,

Wow

Are arbitrary mapping functions, each corresponding to local reconstruction information 504. Also,

Is the spatial activity of the block,

Is the average of the spatial activity of each block.

For example, as the spatial activity of the block increases, the block may be classified into a region having strong detail, and the block may be classified into a region where the detail of the image is weak as the spatial activity of the block is small. At this time, a region having high detail is output as a clear image as the degree of enhancement of the high frequency component is increased, but a problem such as noise boost may occur as a region where the detail is weak is increased as the enhancement degree of the high frequency component is increased.

As a result, in order to prevent the problem, as shown in Equation 2, the restoration parameter extractor 501 may extract a blur value as the spatial activity is large, and extract the blur value as the spatial activity is small. In addition, the restoration parameter extractor 501 may extract the noise removal threshold as a smaller value as the spatial activity is greater, and extract the noise removal threshold as a larger value as the spatial activity is smaller.

As a result, the reconstruction parameter extractor 501 may extract the blur value and the noise removal threshold of each block by using the analysis result of the spatial activity of each block. At this time, the reconstruction parameter extractor 501 extracts reconstruction parameters based on the global reconstruction information 505 for the entire image and the local reconstruction information 504 for each block, thereby preventing blocking artifacts from occurring. .

In this case, the blur value may mean a result of convolution of various blurs, such as blur generated when an image is acquired and blur generated when the image scaler 102 improves the resolution of the image. In this case, when various blurs are convolved, it may be assumed that the blur value is close to the Gaussian function by the central limit theorem. For example, the Gaussian function may be represented by Equation 3 below.

는 블러값을 가우시안 함수 형태로 나타낸 것이고, K는 가우시안 함수를 정규화하기 위한 상수이며,

는 블록 파라미터인 블러값 을 의미한다.

가 클수록, 가우시안 함수

는 전 영역으로 확산된 형태를 나타내기 때문에, 영상의 블러 정도도 높게 나타난다. 반대로,

가 작을수록, 가우시안 함수

는 특정 지점에서 집중된 형태를 나타내기 때문에 영상의 블러 정도는 낮게 나타난다. 일례로, 각 블록의

는 현재 블록의 노이즈와 이전 블록의 노이즈를 이용하여 예측될 수 있다.here,

Represents the blur value in the form of a Gaussian function, K is a constant to normalize the Gaussian function,

Denotes a blur value that is a block parameter.

Is larger, Gaussian function

Since is a form that is spread all over the area, the blur of the image is also high. Contrary,

Is smaller, Gaussian function

Since the image shows a concentrated form at a specific point, the blur of the image is low. For example, in each block

Can be predicted using the noise of the current block and the noise of the previous block.

FIG. 6 is a diagram for describing a process of Fourier transform region filtering performed by an image restoration unit of an image restoration system according to an embodiment of the present invention.

을 이용할 수 있다. 푸리에 변환 영역 필터링(601)은 입력 영상을 푸리에 변환(602)하고, 복원 파라미터(603)인 블록의 블러값을 푸리에 변환(602)할 수 있다.For example, the image reconstructor 105 may apply Fourier transform region filtering 601 to each block using the reconstruction parameter 603. At this time, the Fourier transform region filtering 601 is a blur value among the reconstruction parameters 603.

Can be used. The Fourier transform region filtering 601 may perform a Fourier transform 602 on the input image and perform a Fourier transform 602 on the blur value of the block, which is the reconstruction parameter 603.

와 푸리에 변환(602)된 블러값인

을 이용하여 필터링 영상(604)인

를 결정할 수 있다. 일례로, 영상 복원부(105)는 하기 수학식 4를 통 해 필터링 영상(604)을 결정할 수 있다.As can be seen in Figure 6, the Fourier transform (602) is an input image

And the Fourier transform (602)

Using the filtered image (604)

Can be determined. For example, the image reconstructor 105 may determine the filtered image 604 through Equation 4 below.

,

는 블록의 위치를 나타내며, G는 정규화를 위한 함수를 의미한다. 상기 수학식 4를 참고하면, 고주파일수록 분모 값이 크게 설정되어 역푸리에 변환(605)의 효과를 감소시킨다. 따라서, 필터링 영상(604)인

을 역푸리에 변환(605)할 때 지나친 고주파의 강조를 막을 수 있다. 상기 수학식 4는 일례에 불과하고, 시스템의 구성에 따라 변경될 수 있다.At this time,

,

Denotes the position of the block, and G denotes a function for normalization. Referring to Equation 4, the higher the higher the denominator value is set to reduce the effect of the inverse Fourier transform (605). Therefore, the filtered image 604

When the inverse Fourier transform (605) it can prevent the excessive emphasis of the high frequency. Equation 4 is merely an example and may be changed according to the configuration of the system.

FIG. 7 is a diagram for describing a process of wavelet transform region filtering performed by an image restoration unit of an image restoration system according to an exemplary embodiment of the present invention.

For example, the image reconstructor 105 may apply the wavelet transform region filtering 701 to each block by using the reconstruction parameter 705. For example, the image reconstructor 105 may apply wavelet transform region filtering after Fourier transform region filtering is terminated for each block. However, the image reconstructor 105 may apply the Fourier transform region filtering using the blur value and the wavelet transform region filtering using the noise removal threshold in parallel in consideration of the filtering process for each divided region.

와 복원 파라미터(705)인 노이즈 제거 임계값인

를 이용하여 필터링 영상(703)인

을 결정할 수 있다. 일례로, 영상 복원부(105)는 하기 수학식 5를 통해 필터링 영상(703)을 결정할 수 있다.In this case, the wavelet transform region filtering 701 may use a noise removal threshold value of the recovery parameter 705. As can be seen in Figure 7, the wavelet transform is the input image 702

And the reconstruction parameter 705

Using the filtered image (703)

Can be determined. For example, the image reconstructor 105 may determine the filtered image 703 through Equation 5 below.

를 조절하여 푸리에 변환 영역 필터링에서 발생한 노이즈 부스트가 제거되고, 영상의 선명도가 향상될 수 있다. 이 때, 웨이블릿 계수인

는 로컬 복원 정보(706)에 포함된다. 웨이블릿 변환 영역 필터링(701)은 신호의 특징점(예: 영상의 에지)을 잘 표현할 수 있기 때문에, 노이즈 제거와 선명도 향상에 효율적이다. 이 때, 영상 복원부(105)는 영상이 노이즈로 판단되면 (

), 필터링 영상(703)을 0으로 결정하여 노이즈를 제거하고, 영상이 신호 성분으로 판단되 면(

), 필터링 영상(703)을 웨이블릿 계수인

를 통해 스케일링하여 영상의 선명도를 향상시킬 수 있다. 이 때, 웨이블릿 계수인

는 1보다 큰 값을 가질 수 있다. 필터링 영상(703)이 결정되면, 역 웨이블릿 변환(704)를 통해 노이즈가 제거됨과 동시에 선명도가 향상된 영상이 도출될 수 있다.The image reconstructor 105 is a wavelet coefficient of the wavelet transform region filtering 701.

The noise boost generated by the Fourier transform region filtering may be removed, and the sharpness of the image may be improved. At this time, the wavelet coefficient

Is included in local reconstruction information 706. Since the wavelet transform region filtering 701 can express a feature point of a signal (eg, an edge of an image) well, the wavelet transform region filtering 701 is effective for removing noise and improving sharpness. At this time, the image reconstructor 105 determines that the image is noise (

If the filtered image 703 is determined to be 0 to remove noise, and the image is determined to be a signal component (

), The filtered image 703 is a wavelet coefficient

You can improve the sharpness of the image by scaling through. At this time, the wavelet coefficient

May have a value greater than one. When the filtered image 703 is determined, the image may be removed while the noise is removed through the inverse wavelet transform 704.

8 is a flowchart illustrating an image restoration method according to an embodiment of the present invention.

The image restoration method according to an embodiment of the present invention improves the resolution of the image by upsampling an image and interpolating the upsampled image (S801).

In this case, the method of improving the resolution of an image (S801) may upsample the image using zero data and interpolate the upsampled image through a low band filter.

In the image restoration method according to an embodiment of the present invention, the image boundary is processed by performing color shifting on the boundary region of the image (S802).

At this time, the step of processing the boundary of the image (S802) may process the boundary of the image with improved resolution. The processing of the boundary of the image (S802) may remove the ringing component of the boundary region of the image by shifting the object color and the background color from the boundary of the object color and the background color of the image to respective representative colors. Can be.

The ringing component of the boundary region of the image may be removed by increasing the signal bandwidth of one of the object color and the background color of the image.

In an image restoration method according to an embodiment of the present invention, an image is divided into at least one region, and a restoration parameter is extracted for each block included in the divided region (S803).

At this time, the step of extracting the reconstruction parameter for each block (S803) may extract the blur value and the noise removal threshold value of each block by using the analysis result of the spatial activity of each block.

Specifically, in the step of extracting the reconstruction parameter for each block (S803), a larger blur value can be extracted as the spatial activity of each block is larger, and a blur value can be extracted as the spatial activity is smaller. In addition, in the step of extracting the reconstruction parameter for each block (S803), the noise removal threshold may be extracted as a smaller value as the spatial activity of each block is larger, and the noise removal threshold may be extracted as a larger value as the spatial activity is smaller. .

In this case, in the extracting of the restoration parameter for each block (S803), the restoration parameter may be extracted based on global restoration information for the entire image and local restoration information for each of the blocks.

An image restoration method according to an embodiment of the present invention performs image restoration by applying block-based transform region filtering according to a restoration parameter. In this case, in performing the image restoration, fourier transform region filtering (S804) and wavelet transform region filtering (S805) may be applied to each of the blocks by using the restoration parameters.

In this case, the reconstruction of the image may be performed by applying Fourier transform region filtering (S804) using a blur value and wavelet transform region filtering (S805) using a noise removal threshold in consideration of the filtering process for each divided region. can do.

In addition, the image restoration method according to an embodiment of the present invention includes a computer readable medium including program instructions for performing operations implemented by various computers. The computer readable medium may include program instructions, data files, data structures, etc. alone or in combination. The media may be program instructions that are specially designed and constructed for the present invention or may be available to those skilled in the art of computer software. Examples of computer-readable recording media include magnetic media such as hard disks, floppy disks, and magnetic tape, optical media such as CD-ROMs, DVDs, and magnetic disks, such as floppy disks. Magneto-optical media, and hardware devices specifically configured to store and execute program instructions, such as ROM, RAM, flash memory, and the like. Examples of program instructions include not only machine code generated by a compiler, but also high-level language code that can be executed by a computer using an interpreter or the like.

As described above, the present invention has been described with reference to the limited embodiments and the drawings, but the present invention is not limited to the above embodiments, which can be variously modified by those skilled in the art to which the present invention pertains. And variations are possible. Accordingly, the spirit of the present invention should be understood only in accordance with the following claims, and all equivalents or equivalent variations thereof are included in the scope of the present invention.

1 is a block diagram showing a specific configuration of an image reconstruction system according to an embodiment of the present invention.

2 is a view for explaining the effect of the boundary treatment when restoring a high frequency component according to an embodiment of the present invention.

3 is a view for explaining the operation of the boundary processor of the image restoration system according to an embodiment of the present invention.

4 is a diagram for describing a process of performing block-based image reconstruction according to an embodiment of the present invention.

5 is a diagram for describing a detailed operation of a block parameter extracting unit of an image restoration system according to an embodiment of the present invention.

FIG. 6 is a diagram for describing a process of Fourier transform region filtering performed by an image restoration unit of an image restoration system according to an embodiment of the present invention.

FIG. 7 is a diagram for describing a process of wavelet transform region filtering performed by an image restoration unit of an image restoration system according to an exemplary embodiment of the present invention.

8 is a flowchart illustrating an image restoration method according to an embodiment of the present invention.

<Explanation of symbols for the main parts of the drawings>

101: image restoration system

102: image scaler

103: boundary processing unit

104: restore parameter extraction unit

105: image restoration unit

Claims (21)

A boundary processor configured to process the boundary of the image by performing color shifting on the boundary region of the image; And A reconstruction parameter extractor for dividing the image into at least one area and extracting reconstruction parameters for each block included in the divided area; And An image restoration unit performing image restoration by applying block-based transform region filtering according to the restoration parameters. Image restoration system comprising a. The method of claim 1, An image scaler for upsampling an image and improving the resolution of the image by interpolating the upsampled image More, The boundary processing unit, And a boundary of the image having improved resolution. 3. The method of claim 2, The image scaler unit, Upsampling the image using zero data and interpolating the upsampled image through a low-band filter. The method of claim 1, The boundary processing unit, And a ringing component of the boundary region of the image is removed by shifting the object color and the background color from the boundary between the object color and the background color of the image to each representative color. The method of claim 1, The restoration parameter extraction unit, The image reconstruction system of claim 1, wherein the blur value and the noise removal threshold of each block are extracted by using the analysis result of the spatial activity of each block. The method of claim 5, The restoration parameter extraction unit, The greater the spatial activity is, the larger the blur value is extracted, and the smaller the spatial activity is, the image reconstruction system characterized in that the smaller the blur value is extracted. The method of claim 5, The restoration parameter extraction unit, The noise reduction threshold is extracted as a smaller value as the spatial activity is greater, and the noise removal threshold is extracted as a larger value as the spatial activity is smaller. The method of claim 1, The restoration parameter extraction unit, And a reconstruction parameter is extracted based on global reconstruction information for the entire image and local reconstruction information for each of the blocks. The method of claim 1, The image restoration unit, And Fourier transform region filtering and wavelet transform region filtering are applied to each of the blocks using the reconstruction parameters. 10. The method of claim 9, The image restoration unit, And a Fourier transform region filtering using a blur value and a wavelet transform region filtering using a noise removal threshold in parallel in consideration of the filtering process for each divided region. Processing the boundary of the image by performing color shifting on the boundary region of the image; Dividing the image into at least one area and extracting a restoration parameter for each block included in the divided area; And Performing image reconstruction by applying block-based transform region filtering according to the reconstruction parameter Image restoration method comprising a. 12. The method of claim 11, Upsampling an image and interpolating the upsampled image to improve the resolution of the image More, The step of processing the boundary of the image, And a boundary of the image having improved resolution. The method of claim 12, The step of improving the resolution of the image, And up-sampling the image using zero data and interpolating the upsampled image through a low-band filter. 12. The method of claim 11, The step of processing the boundary of the image, And increasing a signal bandwidth of one of an object color and a background color of the image to remove a ringing component of a boundary region of the image. 12. The method of claim 11, The extracting of the restoration parameter for each block included in the divided region may include: The image reconstruction method of claim 1, wherein the blur value and the noise removal threshold of each block are extracted using analysis results of the spatial activity of each block. 16. The method of claim 15, The extracting of the restoration parameter for each block included in the divided region may include: The greater the spatial activity of each of the blocks, the greater the blur value, and the smaller the spatial activity, the less the blur value, characterized in that the image reconstruction method. 16. The method of claim 15, The extracting of the restoration parameter for each block included in the divided region may include: The noise reduction threshold is extracted as a smaller value as the spatial activity of each block is larger, and the noise removal threshold is extracted as a larger value as the spatial activity is smaller. 12. The method of claim 11, The extracting of the restoration parameter for each block included in the divided region may include: And restoring parameters based on global restoration information on the entire image and local restoration information on each of the blocks. 12. The method of claim 11, The step of performing restoration on the image, And Fourier transform region filtering and wavelet transform region filtering are applied to each of the blocks using the reconstruction parameters. 20. The method of claim 19, The step of performing restoration on the image, And a Fourier transform region filtering using a blur value and a wavelet transform region filtering using a noise removal threshold in parallel in consideration of the filtering process for each divided region. A computer-readable recording medium in which a program for executing the method of any one of claims 11 to 20 is recorded.

Images